Conspicuous Consumption and Satisfaction

This working paper explores the impact of conspicuous consumption on satisfaction and income in Switzerland. It focuses on luxury car consumption and uses household panel data to analyze the relationship between the prevalence of luxury cars in a municipality and individual income satisfaction. The results suggest a negative impact on income satisfaction.

econstor
Make Your Publications Visible.
A Service of
zbwLeibniz-Informationszentrum
Wirtschaft
Leibniz Information Centre
for Economics
Winkelmann, Rainer
Working Paper
Conspicuous consumption and satisfaction
Working Paper, No. 30
Provided in Cooperation with:
Department of Economics, University of Zurich
Suggested Citation: Winkelmann, Rainer (2011) : Conspicuous consumption and satisfaction,
Working Paper, No. 30, University of Zurich, Department of Economics, Zurich,
http://dx.doi.org/10.5167/uzh-51534
This Version is available at:
http://hdl.handle.net/10419/77507
Standard-Nutzungsbedingungen:
Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen
Zwecken und zum Privatgebrauch gespeichert und kopiert werden.
Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle
Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich
machen, vertreiben oder anderweitig nutzen.
Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen
(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten,
gelten abweichend von diesen Nutzungsbedingungen die in der dort
genannten Lizenz gewährten Nutzungsrechte.
Terms of use:
Documents in EconStor may be saved and copied for your
personal and scholarly purposes.
You are not to copy documents for public or commercial
purposes, to exhibit the documents publicly, to make them
publicly available on the internet, or to distribute or otherwise
use the documents in public.
If the documents have been made available under an Open
Content Licence (especially Creative Commons Licences), you
may exercise further usage rights as specified in the indicated
licence.
www.econstor.eu

Conspicuous Consumption and Satisfaction
Rainer Winkelmann ∗
University of Zurich, CESifo and IZA
August 2011
Abstract
Traditionaltools ofwelfare economics identify the envy-related welfare loss from
conspicuous consumption only under very strong assumptions.Measured income and
life satisfaction offers an alternative for estimating such consumption externalities.The
approach is developed in the context of luxury car consumption (Ferraris and Porsches)
in Switzerland.Results from household panel data and fixed effects panel regressions
suggest that the prevalence of luxury cars in the municipality of residence has a negative
impact on own income satisfaction.
JEL Classification:D12, D62, I31
Keywords:Ferrari, Porsche, status, consumption externality, Swiss Household Panel
∗Address for correspondence:University of Zurich,Department of Economics,Zuerichbergstr.14,CH-
8032 Zurich,Switzerland,email:[e-mail].I thank Holger Bonin,Andrew Oswald,
Kevin Staub,two anonymous referees and seminar participants in Bayreuth,Bamberg,Glasgow (Annual
Meetings ofthe European Economic Association 2010,Fraser ofAllander Institute,2011) and Lausanne
(Workshop in Redistribution and Well-Being) for helpfulcomments,Oliver Lipps for support in obtaining
the data, and Wishiro Keo for excellent research assistance.

1 Introduction
There is ample evidence that individualwell-being is affected by comparisons with others.
Status concerns are an important feature of our social existence.Higher ranked individuals
are on average healthier (Wilkinson,2000,Marmot,2003) and live longer (Oswald and
Rablen, 2008) than lower ranked ones.They also report higher levels of happiness in survey
questions (Di Tella at al., 2007).Relatedly, actions of others provide a “frame of reference”
for own decisions (Frank,1991).As Karl Marx (1847) famously noted,“A house may be
large or small;as long as the neighboring houses are likewise small,it satisfies allsocial
requirement for a residence.But let there arise next to the little house a palace,and the
little house shrinks to a hut.”
For the economist, such positional concerns and frame of reference effects are importan
as they can lead to inefficient market outcomes.For example, resources spent on improving
one’s status are wasteful from a societal point of view, as one person’s gain is another pers
loss. As Layard (1980) puts it “For,though individuals are willing to make sacrifices to
improve their individual position, the net result of status-motivated action will be no increas
in status satisfaction but an increase in sacrifice.” (p.738).Perhaps the best studied example
of such wasteful competition occurs in the presence of relative consumption effects, i.e.if an
individual’s utility depends not only on the levelof her own consumption but also on how
that level compares with the consumption of others.
One cause of such interdependence has been explored by Veblen (1899) who, referring
the behavior of the nouveau riche during the Second Industrial Revolution in Britain, coined
the term of “conspicuous consumption”.He argued that this consumption “is evidence of
wealth, and thus becomes honorific, and .. . failure to consume a mark of demerit.” Veblen
thus singles out consumption that is intently used as a signalfor status. More broadly,
1

conspicuous consumption refers to any consumption activity that is, first, literally “visible”
to outsiders,and, second,positional,in the sense that own consumption utility depends
partly on relative rather than absolute consumption.
Despite the obvious interest for microeconomic theory (e.g.Layard,1980,Dupor and
Liu, 2003,Arrow and Dasgupta,2009),and the widespread reference to the concept in
the economic literature, there is surprisingly little direct empirical evidence on the effect of
conspicuous consumption on well-being of others.The objective of this paper is to contribute
such evidence within the context of the life satisfaction literature, in order to provide direct
empirical quantifications of the presence and size of consumption externalities.I propose to
use information on what people say when asked whether they have a good and worthwhile
life, i.e., survey information on “subjective well-being”,“satisfaction”,or “happiness”,in
order to estimate the effect of conspicuous consumption on others’ satisfaction, and thus th
utility loss due to such consumption.The empiricalapproach using satisfaction equations
is in the spirit of previous papers concerned with the valuation of intangibles, including van
Praag and Baarsma (2005) on the cost of airport noise, Luechinger (2009) for air quality an
Frey et al.(2009) for terrorism.
This paper focuses on a specific instance of conspicuous consumption, the purchase and
display of luxury sports cars in Switzerland.Two brands are considered, Ferrari and Porsche.
These are the two main automobile producers present in the Swiss market that specialize
in the upper segment ofluxury sport cars.The demand for these cars has increased over
the recent decade.Whereas totalsales ofnew cars declined between 2001 and 2007,new
registrations of Ferraris and Porsches increased by almost 80% (in 2008,the numbers went
down, see Figure 1).The consequence was a doubling in market share of such luxury sports
2

cars among all newly registered cars from 0.4% in 2001 to 0.8% in 2007.
− − − − −− Figure 1 about here − − − − − − −
The empiricalapproach in this paper combines information from various sources.Indi-
vidual level satisfaction (with income and with life in general) and other socio-demographic
characteristics (most importantly current household income) are obtained from the 2002 an
2007 waves of the Swiss Household Panel.In a second step, regional information is matched
to each person-year observation.The first, obtained from the Federal Roads Office, provides
the number ofnew luxury car registrations (Ferrariand Porsche per 1000 population and
year) in the municipality and canton of residence.1 The second,obtained from the Federal
Tax Administration, includes information on average incomes and income inequality, again
for each municipality, canton and year.
The statistical hypothesis being tested is that income and life satisfaction are not related
to the localor regionaldensity of new luxury sport cars,ceteris paribus,against the alter-
native that there is a negative association.The empirical analysis employs linear regression
models with fixed individual specific as well as region specific effects.
2 Background
2.1 Modeling conspicuous car consumption
The following model,an adaptation of Dupor and Liu (2003),provides a usefulframework
for thinking about conspicuous car consumption.Let c denote own consumption ofcar
attributes (such as horsepower,or looks) and ¯c the average attributes in the population of
1A municipality is the smallest administrative and politicalunit in Switzerland.Depending on size,it
can be a village, town or city.
3

cars.The utility function of identical consumers can be written as
U (c, ¯c, x) (1)
where x is a non-conspicuouscomposite good.Suppose thatpeople enjoy horsepower
(∂U/∂c = U1 > 0) as well as non-conspicuous consumption (∂U/∂x = U3 > 0), that
marginal utility is diminishing (∂2U/∂c2 < 0, ∂2U/∂x 2 < 0), and that horsepower consump-
tion is conspicuous and generates envy (∂U/∂¯c = U2 < 0). People then choose c and x in
order to maximize utility subject to the budget constraint
c + px ≤ yf
where yf is income at full employment and leisure is part of x.Substituting the constraint
into utility function (1), individuals thus maximize
U (c, ¯c, g(c))
where g(c) = (yf − c)/p.They take public consumption ¯c as given and trade off the marginal
utility from own conspicuous consumption with the marginalutility from non-conspicuous
consumption.The first-order condition for a maximum is
U1(c, ¯c, g(c)) + g0
(c)U3(c, ¯c, g(c)) = 0 (2)
or
1
p = U1(c, ¯c, g(c))
U3(c, ¯c, g(c)) (3)
It is a-prioriundetermined whether public consumption ¯c affects the solution to (3),and
hence what the optimal own consumption levels for c and x are.For example, the marginal
rate of substitution between conspicuous and non-conspicuous consumption is independen
¯c if the utility function is either additively separable, or multiplicative (e.g.of Cobb-Douglas
4

form) in ¯c. In these cases, envy has no behavioral consequences although it lowers a person
utility.In other cases, consumption comparisons can lead to behavioral responses.Suppose,
for example,that an increase in ¯c raises the marginalutility ofconspicuous consumption
relative to that of non-conspicuous consumption.As a consequence, it is optimal to reduce
non-conspicuous consumption (including leisure, i.e.work more, for example) and to increase
own conspicuous consumption.In such a case, it is in principle possible to infer the presence
of a consumption externality from observed behavior.In general,behavioraldata cannot
prove the absence of envy effects for the stated reasons.Suppose instead that U (c, ¯c, g(c))
can be measured empirically.Then it becomes possible to determine the effect of ¯c on U , if
any, using statistical methods.
Learning about such envy effects is important from a policy perspective, as they result in
an inefficient allocation in general,and “overconsumption” in particular.To see this,note
that in a symmetric equilibrium with c = ¯c, we obtain
U1(c, c, g(c)) + g0
(c)U3(c, c, g(c)) = 0
This equilibrium is inefficient,since individuals do not account for the negative externality
U2(c, c, g(c)) they impose on others.The social optimum requires that
U1(c, c, g(c)) + U2(c, c, g(c)) + g0
(c)U3(c, c, g(c)) = 0 (4)
Assuming concavity ofU (c, c, g(c)),the optimalcar consumption c∗ that solves (4) is less
than equilibrium consumption,an instance of overconsumption.A government imposing a
luxury tax can remove or alleviate this inefficiency.
The underlying assumption ofidenticalconsumers is admittedly unrealistic,as every
person would then buy cars with similar attributes, whereas in reality Ferraris and Porsches
are only bought by few people.Changing the modelin order to allow for heterogeneity in
5

endowments (whereby “rich” people would have a greater demand for c) would not change
the essential mechanisms leading to overconsumption.
2.2 Empirical literature
Consumption externalities involve two parties:those who generate them and those who
are affected.There are a number ofrecent empiricalstudies regarding the first group of
people, i.e., the decision of, and demand for, conspicuous consumption.Johannsson-Stenman
and Martinsson (2006) study the perceived importance of status concerns in consumption
decisions.Charles et al.(2009) show that Blacks and Hispanics devote larger shares of
their expenditure bundles to visible goods (clothing, jewelry, and cars) than do Whites with
similar incomes.This observation is consistent with a simple consumption model where the
demand for conspicuous goods is a function of group income.Heffetz (2011) uses a signaling-
by-consuming motive to derive and test differential predictions regarding income elasticitie
of visible and non-visible goods.
In this paper, the focus is instead on the second group of people.Traditionally, economists
have analyzed the effects ofconsumption externalities based on revealed preferences,i.e.,
by studying changes in behavior.For example,as shown above,an increase in comparison
consumption can affect the marginal rate of substitution between consumption of conspicu
and non-conspicuous goods,thereby changing observed choices.Specifically,an increase in
reference consumption can lead to an outward shift in labor supply and an increase in own
consumption,an effect known as “Keeping up with the Joneses” (Frank,1999). A well
known study along this line is Neumark and Postlewaite (1998) who show that a woman’s
employment probability increases ifthe sister’s husband’s income is greater than the own
husband’s income.More recently, Kuhn et al.(2011) show that living next to a winner in the
Dutch Postcode Lottery increases the level of own car consumption.Even if an externality
6

leaves the marginalrate ofsubstitution between conspicuous and non-conspicuous goods
unchanged,other margins can be affected,for example,those related to residentialchoice.
Since conspicuous consumption can be interpreted as a negative neighborhood attribute ¯c,
its effect on U can, in principle, be determined using location decisions (or backed out from
housing prices using a hedonic regression).
An alternative to the revealed preference approach is the stated preference method.Stated
preferences,as elicited in discrete choice experiments (DCE),rely on hypotheticalchoices.
The key advantage of DCEs over revealed preference studies is that it gives the researcher
full freedom and controlin manipulating relative positions.In particular,it is possible to
include the consumption level of others into the choice set (thereby effectively internalizing
it). An example ofthis approach is the “paralleluniverse” question,where one has to
choose between two fictitious societies,for example one,society A,where everyone lives in
a house with 3,000 square feet offloor space,and another one,society B,where the own
house has 4,000 square feet while neighbors have 5,000 square feet.Frank (1999,2004)
contents that most people prefer society A.Solnick and Hemenway (1998) have conducted
more elaborate discrete choice experiments with realpopulations.Interestingly,they find
that leisure appears to be inconspicuous – perhaps in part because it is not so easily observ
by others.
The stated preference and the revealed preference approaches both have their weaknes
The validity of discrete choice experiments critically hinges upon people’s cognitive ability
in valuing hypotheticalalternatives as wellas on their truthfulness in responding to such
questions.DCEs generate hypothetical choice situations, which may be very far from respon-
dent’s actualexperiences,and people may decide differently when faced with realchoices.
The revealed preference approach makes strong assumptions regarding the rationality of
agents and the functioning of markets.For example, it is well possible that individuals are
7

not aware of all relevant neighborhood attributes when they make their location decision.For
long-term residents, transaction costs may prohibit a relocation in the presence of negative
consumption externalities, as long as these costs are sufficiently large.
Given the limitations ofthe stated and revealed preference approaches,I propose to
directly estimate the effect of conspicuous consumption on utility,by using stated life- and
income satisfaction as a proxy for utility.Recent evidence suggests that satisfaction can
be measured with a reasonable degree of reliability,sufficient to be able to compare means
over time or across regions, or to use multiple regression analysis to understand the factors
that affect individual satisfaction (Krueger and Schkade, 2008).Moreover, satisfaction taps
into key components ofquality oflife. For example,many studies have found a positive
correlation between satisfaction, or happiness, and longevity.One of the mechanisms seems
to be that satisfaction protects the healthy from becoming sick,by lowering stress,which
in turn enhances the immune system (Veenhoven,2008). Studies have also shown that
individuals who report higher levels of satisfaction with their lives are rated as happier by
their relatives and friends,tend to smile more during socialinteractions,have higher pre-
frontalbrain activity (the part of the brain associated with positive states),and are more
likely to recall positive life events (Layard, 2005).
There is a sizeable literature on the effect of relative income on satisfaction.Studies using
survey data (e.g.,Clark and Oswald,1996,Luttmer,2005,and Dynan and Ravina,2007)
usually find large negative effects of increased reference incomes ceteris paribus, holding o
income constant.This evidence provides an explanation of the so called Easterlin paradox
(Easterlin,1974) whereby large increases in GDP over time have apparently not led to
correspondingly large increases in average satisfaction.
While relative income effects may be related to conspicuous consumption, the connectio
between the two is not obvious.Unfortunately,direct evidence on the effect ofrelative
8

consumption on satisfaction is much more sparse, perhaps due to the absence of appropria
data sets.A notable exception is Hsee et al.(2008) who report on evidence from data
collected in 31 large cities in China.Participants were asked about their jewelry possessions
and their satisfaction with it.The researchers found that residents ofcities with more
expensive jewelry were not any happier on average about their jewelry than residents of ci
with less expensive jewelry (although within each city,people who owned more expensive
jewelry reported greater satisfaction with their jewelry).
Hsee et al. (2008) note that while some consumption experiences (such as ambient
temperature, having social company etc.) are inherently evaluable and often have a natura
scale,others (the size of a diamond,the horsepower of a car) are not,and hence are much
more prone to socialcomparisons.Of course,comparison effects can also relate to past
own consumption.There is evidence that hedonic adaptation plays a major role in some
consumption activities, such as the size of a house or apartment, but not in others, such as
the duration of a commute (Stutzer and Frey,2008).One factor contributing to the speed
of adaptation is uncertainty:unstable or uncertain consumption experiences create longer
lasting happiness than their stable and certain equivalents (Kahneman and Thaler, 1991).
3 Data
The empiricalanalysis of this paper combines data from a number of sources:information
on individual satisfaction and household income is obtained from the Swiss Household Pane
(SHP); car registration data, aggregated to the municipal and cantonal level, were provided
by the FederalRoads Office;and information on average incomes and income inequality,
again for each municipality and canton,originate from the Swiss FederalTax Administra-
tion. Unfortunately,the years for which the data are available do not exactly match.Car
9

registrations are available for the years 2001 and 2007,whereas income inequality data are
available for the years 2003 and 2006.The mismatch is kept at a minimum (plus or mi-
nus one year) by choosing 2002 and 2007 as base years and extracting the corresponding
person-year observations from the SHP.
The SHP is comparable in structure and scope to other European panel household sur-
veys,such as the German Socio-Economic Panelor the British Household Panel,now Un-
derstanding Society.It was started in 1999 and is ongoing at annualfrequency (Budowski
et al., 2001).Importantly, the Swiss Household Panel collects information on the following
question:In general,how satisfied are you with your life if0 means “not at allsatisfied”
and 10 means “completely satisfied”? In addition,there are questions on satisfaction with
a number oflife domains,including satisfaction with income.The SHP also provides in-
formation on many of the control variables typically used in satisfaction regressions.These
include household income, household size, age, marital status, employment status, and gen
der.Depending on year, there are between 4000 and 5000 person observations with compl
information on the relevant variables.
An innovation of this paper is the augmentation of the SHP data with regional informa-
tion on cars,mean incomes and income inequality.Two levels of regionalaggregation are
considered, the municipality and the canton.Switzerland is composed of about 2,900 munic-
ipalities.According to the population census of 2000, they range in size from 24 inhabitants
(Corippo in the Verzasca Vally in Ticino) to 368,875 inhabitants (Zurich).A total of 1,053
municipalities are represented among participants ofthe Swiss Household Panel.Switzer-
land is a confederation of 26 cantons whose role and political functions closely resemble th
of the States within the U.S.; Again, population sizes vary considerably, from a mere 15,199
10

in the canton of Appenzell Innerrhoden to 1.3 million in the canton of Zurich.
− − − − −− Figure 2 about here − − − − − − −
Exposure to conspicuous car consumption is defined by the number of newly registered
Ferrari and Porsche cars per year per 1000 population in the municipality or canton the ind
vidual resides in.2 We refer to this number from now on as “FP-ratio”.In 2001, the FP-ratio
varied between 0 and 7 at the municipal level and between 0.09 (canton of Berne) and 0.65
(canton of Zug) at the cantonal level (See Figure 2).3 Over the whole of Switzerland, a total
of 1,459 Ferrari and Porsche cars was newly registered during the year 2001, correspondin
to an FP-ratio of about 0.20 per 1000 population.
By 2007,the nationwide number ofnew registrations ofFerraris and Porsches had in-
creased substantially.The “least conspicuous” canton (with a rate of0.08) was now the
canton of Glarus.In Zug, again the canton with by far the highest density of such new sport
cars,the number had almost doubled to 1.27.It has to be pointed out that the number of
newly registered cars may not be the best measure for conspicuous car consumption,as it
measures the flow rather than the stock of such cars.Unfortunately, stock data are not kept
on file by the Federal Roads Office.In the steady state, the stock is proportional to the flow.
Income inequality data stem from the Swiss FederalTax Administration (Jeitziner and
Peters,2007).The SFTA publishes Ginicoefficients for allmunicipalities and cantons of
2A third brand of some importance in the class ofluxury sport cars,Maserati,was excluded because
funding constraints precluded the acquisition of this additional data series.
3The 26 cantons of Switzerland are Zurich (zh); Bern (be); Lucerne (lu); Uri (ur); Schwyz (sz); Obwalden
(ow);Nidwalden (nw);Glarus (gl);Zug (zg);Fribourg (fr);Solothurn (so);Basel-City (bs);Basel-Country
(bl); Schaffhausen (sh); Appenzell Ausserhoden (ar); Appenzell Innerrhoden (ai); St.Gall (sg); Grisons (gr);
Argovia (ag); Thurgau (tg); Ticino (to); Vaud (vd); Valais (vs); Neuchatel (ne); Geneva (ge); Jura (ju).
11

Switzerland,as wellas mean income,median income and number of people living in every
municipality, for selected tax years.Every Swiss resident with income has to file an annual
tax return.Taxable income includes income from all sources (mainly labor earnings, interest
and rental income) but excludes social security contributions (retirement and unemployme
insurance).The income ofjointly declaring couples is divided evenly.The resulting Gini
coefficient measures personal income inequality before taxes.This is not ideal,since utility
is usually derived from disposable income, and many of the channels discussed in the prev
ous section (including reference incomes,altruism and uncertainty) are more appropriately
thought ofin terms ofpost-tax income.However,pre- and post-tax Ginicoefficients are
highly correlated.4
The final sample is restricted to survey participants aged between 20 and 70 who provid
valid information for all left- and right-hand side variables in both 2002 and 2007.Since all
regression specifications include individual specific fixed effects, only persons with two yea
of data contribute to the estimation,and allothers can be disregarded.Table 1 shows the
descriptive statistics for these 4048 observations.
− − − − −− Table 1 about here − − − − − − −
We see that mean income satisfaction decreased between 2002 and 2007 from 7.31 to 7
on the 0-10 scale while average life satisfaction decreased from 8.00 to 7.89.These changes
4For example,Feld et al. (2010,Table A2) report a standard deviation for the difference ofpre- and
post-tax Ginicoefficients at the cantonallevelof 0.21,which,given the respective standard deviations of
2.53 and 2.34 of pre- and post-tax Ginis, implies a correlation of 0.99.
12

are statistically significant.5 Among the potentialcontributing factors,the increase in the
FP-ratio – from 0.24 to 0.31 at the municipal level, and from 0.24 to 0.33 at the canton leve
– is the most “conspicuous”, apart from the fact that all respondents are by construction fiv
years older by the time of the second response.Inequality at the canton level also increased
somewhat,as did average income.Whether the correlation between mean satisfaction and
the FP-ratio over time is just that,a correlation,or else an effect with a plausibly causal
interpretation,will be investigated in the next section by means of an extended regression
analysis.
4 Empirical Models and Results
The effect of conspicuous consumption on satisfaction is estimated by exploiting variation
in the density ofFerraris and Porsches across region and time.Are those who live in a
municipality or canton with relatively more such luxury cars less satisfied than others, and
if so, how large is the effect ofconspicuous consumption on well-being?We would like
to estimate the effect “allelse equal”,and therefore use regression analysis to account for
potentialconfounders.Among them,income is perhaps the most important variable to
include,since it is,first, a proxy for own consumption,and,second,a naturalmonetary
comparison scale in order to gauge the magnitude of the consumption externality.Income is
defined as annual disposable household income, i.e., net of taxes and including all transfers
Other controls are the usual individual socio-demographic variables, an income compar-
ison term (an indicator variable that takes the value ‘one’if own income is above mean
5Note that the two samples are not independent, as they represent repeated measurements on the same
individuals and substantialpositive within-person correlation is to be expected.The standard error of the
average change in income satisfaction is 0.043, which leads to a t-statistic of -2.7.
13

regionalincome,and the value ‘zero’else),and finally individualand region specific fixed
effects.A balanced panel is used, with 2024 person-observations in either year.The models
are estimated with and without controlling for the regionalincome Ginicoefficient.It is
clear that income inequality and conspicuous consumption are closely related.People might
have negative feelings about income inequality per se (Alesina et al., 2004, Winkelmann an
Winkelmann, 2010).Without controlling for income inequality, it is hard to say whether the
effect ofFP-ratio captures genuine conspicuous consumption,or inequality aversion more
broadly.By controlling for income inequality,the specific consumption externality can be
isolated in principle.The question is then whether,for a given levelof inequality,people’s
income and life satisfaction depends on the way the money is spent by those in the upper
parts of the income distribution, namely more or less conspicuously.
Table 2 displays results for the municipality based analysis,Table 3 the results for the
canton based analysis.There are two sets of two regressions (with and without control for the
Gini, respectively), the first set for income satisfaction as dependent variable and the secon
set for overalllife satisfaction.Life satisfaction is often conceptualized as an aggregation
of various domain satisfactions, the income domain being one of them (e.g., van Praag and
Ferrer-i-Carbonell,2004).Since largely unobserved variation in allthe different domains
feed into overall life satisfaction, we would expect that the income satisfaction models have
more explanatory power than the life satisfaction models.
− − − − −− Table 2 about here − − − − − − −
Table 2 shows the estimated coefficients.Household income has a statistically significant
effect on both income and life satisfaction.A 10 percent increase in household income is
predicted to increase income satisfaction by 0.055 points,or .7 percent if evaluated at the
mean income satisfaction of7.2. For life satisfaction,the effect is about halfas large.It
14

is customary in this literature to find statistically significant but economically rather small
income elasticities,regardless ofwhether income satisfaction or life satisfaction is used as
dependent variable.
The key parameter ofinterest is the estimated coefficient ofthe FP-ratio variable.In
the model with income satisfaction as dependent variable, the point estimate is −0.15.The
effect is statistically significant at the 10% level.The magnitude and statistical significance
of the estimate is unaffected by the inclusion of the Gini coefficient as an additional control
variable.Thus, one additional Ferrari or Porsche per thousand population (or 17 additional
cars for the municipality size ofthe average person in the sample) is predicted to lower
income satisfaction by about the same amount as a .015/0.55 = 27% reduction in househo
income.One additionalcar represents about the difference between the least conspicuous
and the most conspicuous canton in 2007.For an alternative perspective,take the 2001
figure of0.20 Ferraris and Porsches per thousand inhabitants in Switzerland.According
to the estimates in Table 2,doubling that rate would have an adverse effect on predicted
income satisfaction equal to that of a 5 percent reduction in income.
While the estimated conspicuous consumption parameter is negative in the life satis-
faction equations as well,it is substantially smaller and no longer statistically significant.
Substantialstatistically uncertainty is also present for the other regression parameters,in-
cluding those capturing the effects ofrelative income and household size.This is in part
a reflection ofthe study design:controlling for individualspecific effects is important to
guard against bias from omitted time-invariant confounders and different anchoring of the
subjective response scales across individuals.With two years ofdata,it also cuts the de-
grees of freedom by half,and it eliminates from the estimation sample allpersons that are
observed for a single year only.The remaining sample of 2048 persons is rather small.Since
15

regional effects are included as well, the amount of available information is further reduced6
Given these limitations,it is even more remarkable that a statistically significant effect of
the municipal FP-ratio on income satisfaction can be found.
In a next step,the analysis is repeated for a different levelof spatialaggregation,the
canton level.It is a-priori unclear, in what direction one would expect the results to move,
as there are a number of opposing forces.For example, there is less variation of the FP-ratio
between cantons than between municipalities and and the precision of the estimates can b
expected to go down as a conseqence.On the other hand,since the number ofnew car-
registrations is used as a proxy for the stock of such cars,the measurement error will tend
to be lower for larger geographic units than for smaller ones (where there is a substantial
fraction of municipalities with zero new registrations in a given year).From a substantial
point of view, the spatial dimension of the consumption externality remains ambiguous.In
the cantons,which are the bigger units,inhabitants in the different parts of the units are
unlikely to know of,and respond to,the behavior of each other.While it is obviously the
case that cars are used to drive around, inside and outside the canton, it is also arguably th
case that the exposure to such consumption externalities is the greatest at the locallevel,
and most so in the immediate neighborhood.
− − − − −− Table 3 about here − − − − − − −
Table 3 shows the regression results for the analysis at the cantonal level.The regressions
models and estimation samples are identical to those used for Table 2, except that the FP-
ratio and Ginicoefficient are measured at the levelof the canton,and that canton rather
than municipality specific fixed effects are included in addition to the individual specific fixe
effects.It turns out that the FP-ratio point estimates effects are actually larger than before,
6The regional effects are identified from movers.
16

in absolute value,in particular in the life satisfaction equations.However,standard errors
have grown much so that none of the negative point estimates is statistically significant.
5 Conclusions
The first objective of this paper was to explore the life satisfaction approach as a conceptua
framework for thinking about conspicuous consumption and consumption externalities.The
traditionaltools ofwelfare analysis identify externalities from behavioralresponses.This
limits the scope ofquestions one can answer.It is for example difficult to quantify the
welfare loss resulting from envy, or to quantify the disutility of an externality, or the utility
of a public good, more broadly.To value such intangibles, the satisfaction approach offers a
potential alternative.
The second objective was to conduct a substantive empiricalanalysis for Switzerland
and to determine the welfare costs of luxury sport cars.The results indicated that people
living in a municipality with a higher prevalence of luxury cars have indeed a lower income
satisfaction than others,ceteris paribus.Doubling the number ofFerraris and Porsche at
the national average is estimated to have an adverse effect on mean income satisfaction th
equals that of a 5 percent reduction in income.The results for life satisfaction and cantonal
spillovers were inconclusive, although negative point estimates were found as well.
Of course,areas with many Ferraris and Porsches are also likely to have more buyers
of other expensive cars,and a higher incidence of other displays of wealth (such as luxury
condominiums).Without explicitly controlling for these factors,the Ferrariand Porsche
density can best be interpreted as a proxy for the overall amount of conspicuous consumpt
that is taking place.The basic assumption is that the more Ferraris and Porsches there are in
a municipality or canton, the higher the incidence of conspicuous consumption, whatever it
17

specific manifestation.The effects should thus not be interpreted as ceteris paribus (keeping
other forms of such conspicuous consumption constant) but rather mutatis mutandis.
Future work in this area should aim at improving statisticalpower.In order to obtain
more precise estimates and make further progress on the issue,one might use more data
of the same kind.For example,purchasing additionalcar registration data would allow
extending the analysis to other years.Or else,one might look for related but different
data.In 2009, an “initiative for human friendly cars” was successful in collecting a sufficient
number of signatures,so that a referendum proposing restrictions on the type of cars sold
as wellas additionalluxury taxes for certain classes of vehicles must be put to the people
in Switzerland in due course.Once available,such voting results can be used to confirm
whether the proposed policy measures find more support in regions where the consumption
externalities are objectively more present.
18

6 References
Alesina, A., R. Di Tella and R. MacCulloch (2004) Inequality and happiness:are Europeans
and Americans different? Journal of Public Economics 88, 2009-2042.
Arrow,K.J. and P.S.Dasgupta (2009) Conspicuous Consumption,Inconspicuous Leisure,
Economic Journal 119:F497-F516.
Budowski,M.,R. Tillmann,E. Zimmermann,B. Wernli,A. Scherpenzeeland A. Gabad-
inho (2001).The Swiss Household Panel 1999-2003:Data for research on microsocial
change, ZUMA-Nachrichten.50:100-125.
Charles,K.K., E. Hurst and N.Roussanov (2009) Conspicuous Consumption and Race,
Quarterly Journal of Economics, 124, 425-467.
Clark, A.E. and A.J. Oswald (1996) Satisfaction and comparison income, Journal of Public
Economics, 61, 359-381.
Di Tella, R., J. Haisken-De New and R. MacCulloch (2007) Happiness Adaptation to Income
and to Status in an Individual Panel, NBER Working Papers 13159.
Dupor,B., and W.-F. Liu (2003) Jealousy and Equilibrium Overconsumption,American
Economic Review, 93, 423-428.
Dynan,K.E., and E. Ravina (2007) Increasing Income Inequality,ExternalHabits,and
Self-Reported Happiness, American Economic Review, 97, 226-231.
Easterlin,Richard A. (1974) Does Economic Growth Improve the Human Lot?Some
Empirical Evidence.In Nations and Households in Economic Growth:Essays in Honor
of Moses Abramovitz, ed.Paul A. David and Melvin W. Reder.New York:Academic
Press, 89-125.
19

Feld, L.P., J. Fischer and G. Kirchg¨assner (2010) The effect of direct democracy on income
redistribution:evidence for Switzerland, Economic Inquiry, 48, 817-840.
Frank, R.H. (1991) Positional externalities, In:Zeckhauser, R.J. (Ed.), Strategy and Choice,
MIT Press, Cambridge, MA, 25-47.
Frank, R. (1999) Luxury fever:Why money fails to satisfy in an era of excess.New York:
Free Press.
Frank, R. (2004) How not to buy happiness, Deadalus, 133(2), 69-79.
Frey,B.S., S. Luechinger and A.Stutzer (2009) The life satisfaction approach to valuing
public goods:The case of terrorism, Public Choice, 138, 317-345.
Heffetz, O. (2011) A Test of Conspicuous Consumption:Visibility and Income Elasticities,
Review of Economics and Statistics, forthcoming.
Hsee, C.K., F. Xu and N. Tang (2008) Two Recommendations on the Pursuit of Happiness,
Journal of Legal Studies, 37, S115- S132.
Jeitziner, B. and R. Peters (2007) Regionale Einkommens- und Verm¨ogensverteilung in der
Schweiz:Was sagen die Steuerdaten? Die Volkswirtschaft, issue 12.
Johansson-Stenman, O. and P. Martinsson (2006) Honestly, why are you driving a BMW?
Journal of Economic Behavior & Organization, 60, 129-146.
Kahneman,D. and R.H.Thaler (1991) Economic Analysis and the Psychology of Utility:
Applications to Compensation Policy, American Economic Review, 81, 341-346.
Krueger, A.B. and D.A. Schkade (2008) The Reliability of Subjective Well-Being Measures,
Journal of Public Economics, 92, 1833-1845.
20

Kuhn, P., P. Kooreman,A.R. Soetevent and A.Kapteyn (2011) The Effects ofLottery
Prizes on Winners and their Neighbors:Evidence from the Dutch Postcode Lottery,
forthcoming in:American Economic Review.
Layard, R. (1980) Human Satisfactions and Public Policy, Economic Journal, 90, 737-750.
Layard, R. (2005) Happiness - Lessons from a New Science, Penguin Press, New York.
Luechinger, S. (2009) Valuing Air Quality Using the Life Satisfaction Approach, Economic
Journal, 119, 482-515.
Luttmer, E.F.P. (2005) Neighbors as negatives:relative earnings and well being, Quarterly
Journal of Economics, 120, 963-1002.
Marmot,M. (2003) Understanding SocialInequalities in Health,Perspectives in Biology
and Medicine 46, S9-S23.
Marx, K. (1987[1847]) Wage, Labour and Capital.
Neumark,D. and A. Postelwaite (1998) Relative income concerns and the rise in married
women’s employment, Journal of Public Economics, 70, 157-193.
Oswald, A.J. and M.D. Rablen (2008) Mortality and Immortality:The Nobel Prize as an
Experiment into the Effect ofStatus upon Longevity,Journal of Health Economics,
27, 1462-1471.
SFSO (2005) Eidgen¨ossische Volksz¨ahlung 2000:Die Raumgliederung der Schweiz, Neuen-
burg:Swiss Federal Statistical Office.
Solnick, S. and D. Hemenway (1998) Is more always better?:a survey on positional concerns.
Journal of Economic Behavior and Organization, 37, 373-383.
21

Stutzer A. and B.S. Frey (2008) Stress that Doesn’t Pay:The Commuting Paradox, Scan-
dinavian Journal of Economics, 110, 339-366.
van Praag, B.M.S. and B.E. Baarsma (2005) Using Happiness Surveys to Value Intangibles:
The Case of Airport Noise, Economic Journal, 115, 224-246.
van Praag, B. and A. Ferrer-i-Carbonell (2004) Happiness Quantified - A Satisfaction Cal-
culus Approach, Oxford University Press.
Veblen,T. (1925[1899]) The Theory of the Leisure Class:An Economic Study of Institu-
tions.
Veenhoven,R. (2008) Healthy happiness:effects of happiness on physicalhealth and the
consequences for preventive health care, Journal of Happiness Studies, 9, 449-469.
Wilkinson, R. (2000) Mind the Gap:Hierarchies, Health and Human Evolution, Weidenfeld
& Nicolson, London.
Winkelmann L. and R. Winkelmann (2010) Does inequality harm the middle class? Kyklos,
63, 301-316.
22

Table 1
Descriptive Statistics
Year 2002 Year 2007
mean std.err. mean std.err.
Income Satisfaction 7.31 (0.041) 7.19 (0.044)
Life Satisfaction 8.00 (0.031) 7.89 (0.030)
Household size 3.08 (0.032) 2.93 (0.031)
Log household income 10.88 (0.011) 10.98 (0.011)
Married (Yes=1) 0.70 (0.010) 0.71 (0.010)
Ferrari&Porsche/1000 pop.(munic.) 0.24 (0.008) 0.31 (0.016)
Ferrari&Porsche/1000 pop.(canton) 0.24 (0.002) 0.33 (0.005)
Gini (munic.) 0.29 (0.001) 0.29 (0.001)
Gini (canton) 0.30 (0.001) 0.32 (0.001)
Log Mean income (munic.) 10.84 (0.004) 10.89 (0.004)
Log Mean income (canton) 10.86 (0.002) 10.91 (0.003)
Number of observations 2024 2024
23

60
80
100
120
140
160
180
2000 2002 2004 2006 2008
Year
Porsche&Ferrari All
Source: www.auto−schweiz.ch
Fig. 1. Index of New Car Registrations in Switzerland (2000=100)
0
.2
.4
.6
Newly registered Ferrari & Porsche
be lu ur ar ju bs so fr ow gr gl ne sg vs sh tg ag ai bl nwvd zh ge ti sz zg
Source: Federal Roads Office, own calculations.
Fig. 2. Cantonal ranking in luxury car registrations per 1000 pop., 2001
24

Table 2
Satisfaction, Inequality and Luxury Cars in Municipality
Income SatisfactionLife Satisfaction
Log household income 0.546** 0.548** 0.287** 0.285**
(0.142) (0.142) (0.105) (0.105)
Income above municipal mean0.129 0.127 0.035 0.036
(0.116) (0.116) (0.086) (0.086)
Ferrari&Porsche/1000 pop. -0.150* -0.147* -0.045 -0.047
(0.082) (0.082) (0.061) (0.061)
Municipal Gini -0.015 0.012
(0.024) (0.018)
Log household size 0.118 0.115 0.024 0.026
(0.177) (0.177) (0.131) (0.131)
R-squared 0.220 0.221 0.162 0.162
Individual fixed effects yes yes yes yes
Municipality fixed effects yes yes yes yes
Notes:
Number of observations:4048;Source:Swiss Household Panel;Other controlvariables not shown in table
are a second order polynomial in age, marital status, and unemployment; standard errors in parentheses; *
p< 0.10, ** p< 0.05
25

Table 3
Satisfaction, Inequality and Luxury Cars in Canton
Income SatisfactionLife Satisfaction
Log household income 0.326** 0.318** 0.242** 0.247**
(0.127) (0.127) (0.092) (0.092)
Income above canton mean0.320** 0.327** -0.020 -0.024
(0.106) (0.106) (0.077) (0.077)
Ferrari&Porsche/1000 pop.-0.310 -0.046 -0.140 -0.294
(0.343) (0.385) (0.248) (0.278)
Cantonal Gini -0.049 0.028
(0.032) (0.023)
Log household size 0.072 0.073 0.055 0.054
(0.136) (0.136) (0.098) (0.098)
R-squared 0.050 0.051 0.029 0.029
Individual fixed effects yes yes yes yes
Canton fixed effects yes yes yes yes
Notes:see Table 2.
26